Frosty's auto-trading bot goes live with REAL money

But that is the advice I gave you, when to shut it down:

when on big trend days your bot loses money. You say you use a long and short term trend to find entry signals. If both long and short are trending strongly I would want my system to get me in long or short in that direction....

Quote from frostengine:

how to know when to shut it down, how to know when to tweak and what tweaks to think about..
More...
 
Quote from walterjennings:

i think i was misunderstood when i said 'do the opposite'. i
backtesting is an easy way to see if the systems compliment each other or if B just reduces the expected profit of running purely A. since frost runs a counter trend system, it may be advantageous to start designing/testing a trend system which compliments it during periods where capital is not being utilized. having the second system may not only produce better capital utilization, but also reduce drawdowns of the counter trend system running alone, hopefully.
More...
Exactly.....use backtesting here, not Markowitz unless you've got dozens of systems in the mix. To add to the sophistication, employe some sort of allocation algorithm to determine the percentages of capital each system is given.
 
Quote from bolter:

kohanz,

You make an important point that is widely misunderstood - you should seek uncorrelated systems (ie: =0). Two systems with a correlation of -1 will tend to generate returns that approaches 0 over time.

bolter
More...

That is correct. I don't know why I kept repeating negative correlation, probably because I've heard it used around here in this same context a lot. My engineering instinct was telling me that uncorrelated is the goal, but I chickened out ;)
 
optioncoach,

The description of my strategy shows why just because its a strong trend day does not necessarily mean it will make money.... the bot does NOT start trading until 9:00 central... the vast majority of the days where it does not make money on trend days is when 80% of the move is ALREADY done by time the bot gets the green light to trade... not only that but the bot has to wait for a signal, on a really strong trend day the bot has to wait for the short trend to start diverging from the longer trend.... sometimes on those really strong trend days this does not happen until a lot of the move is already banked..... By no means is this strategy perfect, but it does have an edge. no reason to shut it down just because it misses out on those days.
 
Hmm, good points. Looking back I made some assumptions based on our bot that would not apply in other scenarios.

I would think that 40% vs 50% is almost identical, I was talking about using a 10% system to pad a 50% system and saying that the risk of cannibalization was too high.

I suppose in backtesting running bot at the same time (to the exclusion of each other) would reveal whether they were interfering with each other.

I was also assuming that the bot was very active and took positions almost every day, it's rare that our system doesn't make a trade on any given day so that would be different for other systems.

However, that being said, I don't believe there is any reliable method (ie > 80%) to discern whether any day will be choppy or not. If both bots are similar in profits and drawdown then you would have to run them in the same logic (or at least allow them to compete) to make sure that one wasn't making money to the exclusion of the other. Two bots can make money side by side independently, but perhaps not when they have to share limited buying power.

In an ideal situation one would take a position and not the other, but what if a day is choppy and becomes trendy? or vice versa? It happens often. If the machine takes a position with one algo, but then could take a BETTER position later on with the other, 1 + 1 = .75

Backtesting would be the key.
 
I agree that locking into a single bot for the whole day would be a mistake, it would have to be dynamic enough to be able to switch back and forth depending on whats happening, or better the bots should do that themselves.

You have to view both bots as a single system i think. even running a single system you'll run into issues where it will take a position and a better position will come along after. the key is to make the best decisions at any giving point based on the known information.
 
+$50 today.....

The bot only traded 3 days this week, for a profit of +$105 for the week.

From the time the bot went live to now, it has traded 29 days for a loss of -$543 if you include my -$744 manual screwup day.

This means the bot has lost an average of: about $19 per day.

If you do not include the -$744 manual mistake and use the -$200 which is what it would have been otherwise then the bot is up +$1 since going live 29 trading days ago.

I expect in a few weeks the markets will begin trading in a more favorable fashion.... So I expect to be net positive by the end of January.....
 
Quote from frostengine:

+$50 today.....

The bot only traded 3 days this week, for a profit of +$105 for the week.

From the time the bot went live to now, it has traded 29 days for a loss of -$543 if you include my -$744 manual screwup day.

This means the bot has lost an average of: about $19 per day.

If you do not include the -$744 manual mistake and use the -$200 which is what it would have been otherwise then the bot is up +$1 since going live 29 trading days ago.

I expect in a few weeks the markets will begin trading in a more favorable fashion.... So I expect to be net positive by the end of January.....
More...

Frost,

Congratulations on following through to the new year. Lately the volume has bee n a little thin for the number of contracts I want to trade in ER2, and I also expect the action to pick up in the new year (probably to the downside). Good luck and happy new year,

RoughTrader
 
Quote from frostengine:

+$50 today.....

The bot only traded 3 days this week, for a profit of +$105 for the week.

From the time the bot went live to now, it has traded 29 days for a loss of -$543 if you include my -$744 manual screwup day.

This means the bot has lost an average of: about $19 per day.

If you do not include the -$744 manual mistake and use the -$200 which is what it would have been otherwise then the bot is up +$1 since going live 29 trading days ago.

I expect in a few weeks the markets will begin trading in a more favorable fashion.... So I expect to be net positive by the end of January.....
More...

You expect the market to adjust to your trading parameters Frost. That is an expectation which is bordering on the delusional. The markets have pretty much been trading in their usual fashion for the past month. If you're bot can't capitalize on these trend days, it's going to be really, really, really tough to actually make money with this thing.

The minute you start to adjust your thinking to take advantage of the opportunities which the markets are providing us every day to trade sucessfully, and stop trying to tell the markets what to do, you will have turned a huge corner in your trading career.

Happy New Year,

JJ

P.S. Today's action was about as pure a trend (in this case, down) as we're ever gonna see.
 

Attachments

Quote from JimmyJam:

P.S. Today's action was about as pure a trend (in this case, down) as we're ever gonna see. [/B]
More...
Exactly. After 10 am EST, it was nearly straight down. It can't get, and it won't get much better than this unless we go nuclear ala October 1987. Astute traders who went short at 10 am were "up" at least $700 per contract by the close.
 
You must log in or register to reply here.
Back
Top